[Madwifi-devel] Data Rate 0 Mb/s (ahdemo, 11a, sample rate adaptation algorithm)

[Mary <mar...@gm...>]

Hello everybody,
I'm doing some experiments with AR5212 cards in ahdemo mode, with
madwifi 0.9.2.1.
I'm using a modified version of sample.c that writes detailed
information about the sent packets on a proc file.
Meanwhile I capture these same packets with another host in monitor mode.

Experiments setting:
 * two hosts with one minipci with the AR5212 chip each are sending
1pkt per second of a fixed size one to each other.
  * a laptop with a AR5212 chip is monitoring and capturing the
exchanged packets
  * ahdemo, 11a, rate adaptation enable (sample rate)

The detailed output that I can read in the proc file system contains
the values of xmit_rate0, xmit_rate1, xmit_rate2, xmit_rate3 (of
struct ar5212_desc) and the respective xmit_tries0, xmit_tries1,
xmit_tries2, xmit_tries3, after that the card consider ended the
sending of a packet (so when ath_rate_tx_complete() in sample.c is
called).

In this output I can notice packets for which xmit_rate1, and
xmit_rate2 are 0 while
xmit_tries1 and xmit_tries2 are not zero.

Looking at the capture I can find this same packets and in the Prism
Monitoring Header, the field Data Rate reports 0.0 Mb/s

And... these packets are reported as successful (ds_txstat.ts_status
of struct ath_desc)

The fact that the monitor report 0Mbps makes me think that 0Mbps is
the value written in the PLCP header... but in this case I'm wondering
how the receiver can decode and ack the packet (since is reported as
succesful...)

Any ideas of where the problem is generated and what is actually happening?!


Thanks a lot,

-- 
___Mary___